How to output "Trace" list from auto.arima in forecast library

Take a look at ?sink

The trace is just a series of calls to 'cat', so it won't be
particularly useful unless you parse.

Jeff

On Mon, Oct 3, 2011 at 2:18 AM, Arun Krishnamoorthy
Hi,
I'm working on the forecast library and am using the auto.arima function
with some dummy data

Fit<-auto.arima(data_ts, trace=TRUE)

I understand that trace evaluates alternative models and provides the
corresponding AIC/SIC values with an output that looks like this;

ARIMA(2,1,2) with drift ? ? ? ? : 1278.988
?ARIMA(0,1,0) with drift ? ? ? ? : 1281.161
?ARIMA(1,1,0) with drift ? ? ? ? : 1280.325
?ARIMA(0,1,1) with drift ? ? ? ? : 1278.2
?ARIMA(1,1,1) with drift ? ? ? ? : 1280.229
?ARIMA(0,1,2) with drift ? ? ? ? : 1278.824
?ARIMA(1,1,2) with drift ? ? ? ? : 1281.911
?ARIMA(0,1,1) ? ? ? ? ? ? ? ? ? ?: 1277.73
?ARIMA(1,1,1) ? ? ? ? ? ? ? ? ? ?: 1279.804
?ARIMA(0,1,0) ? ? ? ? ? ? ? ? ? ?: 1281.265
?ARIMA(0,1,2) ? ? ? ? ? ? ? ? ? ?: 1278.626
?ARIMA(1,1,2) ? ? ? ? ? ? ? ? ? ?: 1281.64

?Best model: ARIMA(0,1,1)

Unfortunately, Trace is not an object and i'm unable to read it in to a data
frame to do further diagnosis.

Can someone please help me with how i can get this output into a data frame.
I'm unable to do data.frame since i can't coerce an ARIMA object into this
class.

Sorry if this has been addressed before. I just think there may be some data
series where running a closely competing alternative model may be useful

Thanks,
AK

-----Original Message-----
From: r-sig-finance-bounces at r-project.org
[mailto:r-sig-finance-bounces at r-project.org] On Behalf Of Daniel Cegielka
Sent: 26 September 2011 17:20
To: chrisbird
Cc: r-sig-finance at r-project.org
Subject: Re: [R-SIG-Finance] Filtering dates/times from zoo/xts series

2011/9/26 chrisbird <chris at chrisbird.com>

Thanks Brian,

I did try using the ['T09:00/T21:00'] method for extraction but it did not
return anything - I will reinvestigate this and see if I can get it
working.

It works... you are sure that the data was ok? Very strange that you have
received nothing...

d<-xts(1:25, Sys.time() + 1:25)> d ? ? ? ? ? ? ? ? ? ?[,1]
2011-09-26 13:26:55 ? ?1
2011-09-26 13:26:56 ? ?2
2011-09-26 13:26:57 ? ?3
2011-09-26 13:26:58 ? ?4
2011-09-26 13:26:59 ? ?5
2011-09-26 13:27:00 ? ?6
2011-09-26 13:27:01 ? ?7
2011-09-26 13:27:02 ? ?8
2011-09-26 13:27:03 ? ?9
2011-09-26 13:27:04 ? 10
2011-09-26 13:27:05 ? 11
2011-09-26 13:27:06 ? 12
2011-09-26 13:27:07 ? 13
2011-09-26 13:27:08 ? 14
2011-09-26 13:27:09 ? 15
2011-09-26 13:27:10 ? 16
2011-09-26 13:27:11 ? 17
2011-09-26 13:27:12 ? 18
2011-09-26 13:27:13 ? 19
2011-09-26 13:27:14 ? 20
2011-09-26 13:27:15 ? 21
2011-09-26 13:27:16 ? 22
2011-09-26 13:27:17 ? 23
2011-09-26 13:27:18 ? 24
2011-09-26 13:27:19 ? 25> d["2011-09-26 13:27:00/2011-09-26 13:27:10"]
? ? ? ? ? ? ? ? ? [,1]
2011-09-26 13:27:00 ? ?6
2011-09-26 13:27:01 ? ?7
2011-09-26 13:27:02 ? ?8
2011-09-26 13:27:03 ? ?9
2011-09-26 13:27:04 ? 10
2011-09-26 13:27:05 ? 11
2011-09-26 13:27:06 ? 12
2011-09-26 13:27:07 ? 13
2011-09-26 13:27:08 ? 14
2011-09-26 13:27:09 ? 15
2011-09-26 13:27:10 ? 16> d["T13:27:00/T13:27:10"] ? ? ? ? ? ? ? ? ? ?[,1]
2011-09-26 13:27:00 ? ?6
2011-09-26 13:27:01 ? ?7
2011-09-26 13:27:02 ? ?8
2011-09-26 13:27:03 ? ?9
2011-09-26 13:27:04 ? 10
2011-09-26 13:27:05 ? 11
2011-09-26 13:27:06 ? 12
2011-09-26 13:27:07 ? 13
2011-09-26 13:27:08 ? 14
2011-09-26 13:27:09 ? 15
2011-09-26 13:27:10 ? 16

The processing is not to remove non-trading days/holidays - I do that
elsewhere. I'm doing processing on some complex strategies which use some
instruments which trade a lot, but not everyday. I only wish to process
the
data from liquid days and strip out the less liquid data.

It's quite a sophisticated approach to data. Probably when you filter using
the time you would have to count the number of observations and does not
bind data below a certain level. I have no idea how to do it in an elegant
way.

best regards,
daniel

I will certainly investigate quantstrat.

Thanks,

Chris.

--
View this message in context:

http://r.789695.n4.nabble.com/Filtering-dates-times-from-zoo-xts-series-tp38
42937p3843408.html
Sent from the Rmetrics mailing list archive at Nabble.com.

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions
should go.

? ? ? ?[[alternative HTML version deleted]]

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions
should go.

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions should go.

Jeffrey Ryan
jeffrey.ryan at lemnica.com

www.lemnica.com
www.esotericR.com
In addition to sink(), you can also use capture.output().  Then you
can read the contents of "out" into a new object.

out <- capture.output({
  Fit <- auto.arima(data_ts, trace=TRUE)
})
outData <- read.table(con <- textConnection(out), sep=":")
close(con)

Best,
--
Joshua Ulrich ?| ?FOSS Trading: www.fosstrading.com
Take a look at ?sink

The trace is just a series of calls to 'cat', so it won't be
particularly useful unless you parse.

Jeff

On Mon, Oct 3, 2011 at 2:18 AM, Arun Krishnamoorthy
<arun.k at bridgei2i.com> wrote:
Hi,
I'm working on the forecast library and am using the auto.arima function
with some dummy data

Fit<-auto.arima(data_ts, trace=TRUE)

I understand that trace evaluates alternative models and provides the
corresponding AIC/SIC values with an output that looks like this;

ARIMA(2,1,2) with drift ? ? ? ? : 1278.988
?ARIMA(0,1,0) with drift ? ? ? ? : 1281.161
?ARIMA(1,1,0) with drift ? ? ? ? : 1280.325
?ARIMA(0,1,1) with drift ? ? ? ? : 1278.2
?ARIMA(1,1,1) with drift ? ? ? ? : 1280.229
?ARIMA(0,1,2) with drift ? ? ? ? : 1278.824
?ARIMA(1,1,2) with drift ? ? ? ? : 1281.911
?ARIMA(0,1,1) ? ? ? ? ? ? ? ? ? ?: 1277.73
?ARIMA(1,1,1) ? ? ? ? ? ? ? ? ? ?: 1279.804
?ARIMA(0,1,0) ? ? ? ? ? ? ? ? ? ?: 1281.265
?ARIMA(0,1,2) ? ? ? ? ? ? ? ? ? ?: 1278.626
?ARIMA(1,1,2) ? ? ? ? ? ? ? ? ? ?: 1281.64

?Best model: ARIMA(0,1,1)

Unfortunately, Trace is not an object and i'm unable to read it in to a data
frame to do further diagnosis.

Can someone please help me with how i can get this output into a data frame.
I'm unable to do data.frame since i can't coerce an ARIMA object into this
class.

Sorry if this has been addressed before. I just think there may be some data
series where running a closely competing alternative model may be useful

Thanks,
AK

-----Original Message-----
From: r-sig-finance-bounces at r-project.org
[mailto:r-sig-finance-bounces at r-project.org] On Behalf Of Daniel Cegielka
Sent: 26 September 2011 17:20
To: chrisbird
Cc: r-sig-finance at r-project.org
Subject: Re: [R-SIG-Finance] Filtering dates/times from zoo/xts series

2011/9/26 chrisbird <chris at chrisbird.com>

Thanks Brian,

I did try using the ['T09:00/T21:00'] method for extraction but it did not
return anything - I will reinvestigate this and see if I can get it
working.

It works... you are sure that the data was ok? Very strange that you have
received nothing...

d<-xts(1:25, Sys.time() + 1:25)> d ? ? ? ? ? ? ? ? ? ?[,1]
2011-09-26 13:26:55 ? ?1
2011-09-26 13:26:56 ? ?2
2011-09-26 13:26:57 ? ?3
2011-09-26 13:26:58 ? ?4
2011-09-26 13:26:59 ? ?5
2011-09-26 13:27:00 ? ?6
2011-09-26 13:27:01 ? ?7
2011-09-26 13:27:02 ? ?8
2011-09-26 13:27:03 ? ?9
2011-09-26 13:27:04 ? 10
2011-09-26 13:27:05 ? 11
2011-09-26 13:27:06 ? 12
2011-09-26 13:27:07 ? 13
2011-09-26 13:27:08 ? 14
2011-09-26 13:27:09 ? 15
2011-09-26 13:27:10 ? 16
2011-09-26 13:27:11 ? 17
2011-09-26 13:27:12 ? 18
2011-09-26 13:27:13 ? 19
2011-09-26 13:27:14 ? 20
2011-09-26 13:27:15 ? 21
2011-09-26 13:27:16 ? 22
2011-09-26 13:27:17 ? 23
2011-09-26 13:27:18 ? 24
2011-09-26 13:27:19 ? 25> d["2011-09-26 13:27:00/2011-09-26 13:27:10"]
? ? ? ? ? ? ? ? ? [,1]
2011-09-26 13:27:00 ? ?6
2011-09-26 13:27:01 ? ?7
2011-09-26 13:27:02 ? ?8
2011-09-26 13:27:03 ? ?9
2011-09-26 13:27:04 ? 10
2011-09-26 13:27:05 ? 11
2011-09-26 13:27:06 ? 12
2011-09-26 13:27:07 ? 13
2011-09-26 13:27:08 ? 14
2011-09-26 13:27:09 ? 15
2011-09-26 13:27:10 ? 16> d["T13:27:00/T13:27:10"] ? ? ? ? ? ? ? ? ? ?[,1]
2011-09-26 13:27:00 ? ?6
2011-09-26 13:27:01 ? ?7
2011-09-26 13:27:02 ? ?8
2011-09-26 13:27:03 ? ?9
2011-09-26 13:27:04 ? 10
2011-09-26 13:27:05 ? 11
2011-09-26 13:27:06 ? 12
2011-09-26 13:27:07 ? 13
2011-09-26 13:27:08 ? 14
2011-09-26 13:27:09 ? 15
2011-09-26 13:27:10 ? 16

The processing is not to remove non-trading days/holidays - I do that
elsewhere. I'm doing processing on some complex strategies which use some
instruments which trade a lot, but not everyday. I only wish to process
the
data from liquid days and strip out the less liquid data.

It's quite a sophisticated approach to data. Probably when you filter using
the time you would have to count the number of observations and does not
bind data below a certain level. I have no idea how to do it in an elegant
way.

best regards,
daniel

I will certainly investigate quantstrat.

Thanks,

Chris.

--
View this message in context:

http://r.789695.n4.nabble.com/Filtering-dates-times-from-zoo-xts-series-tp38
42937p3843408.html
Sent from the Rmetrics mailing list archive at Nabble.com.

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions
should go.

? ? ? ?[[alternative HTML version deleted]]

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions
should go.

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions should go.

--
Jeffrey Ryan
jeffrey.ryan at lemnica.com

www.lemnica.com
www.esotericR.com

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions should go.

Works like magic!! Thanks Jeff and Joshua. 

There is a clear advantage in running lower order ARIMA models with drift in
comparison to what auto.arima gives. The AIC values just differ by a few
points so there isn't really much to choose based just on AIC. The competing
model however offers significant reduction in prediction error. 

Thanks for the help again.

-----Original Message-----
From: Joshua Ulrich [mailto:josh.m.ulrich at gmail.com] 
Sent: 03 October 2011 19:58
To: Jeffrey Ryan
Cc: Arun Krishnamoorthy; r-sig-finance at r-project.org
Subject: Re: [R-SIG-Finance] How to output "Trace" list from auto.arima in
forecast library

In addition to sink(), you can also use capture.output().  Then you
can read the contents of "out" into a new object.

out <- capture.output({
  Fit <- auto.arima(data_ts, trace=TRUE)
})
outData <- read.table(con <- textConnection(out), sep=":")
close(con)

Best,
--
Joshua Ulrich ?| ?FOSS Trading: www.fosstrading.com

On Mon, Oct 3, 2011 at 7:36 AM, Jeffrey Ryan <jeffrey.ryan at lemnica.com>
wrote:
Take a look at ?sink

The trace is just a series of calls to 'cat', so it won't be
particularly useful unless you parse.

Jeff

On Mon, Oct 3, 2011 at 2:18 AM, Arun Krishnamoorthy
<arun.k at bridgei2i.com> wrote:
Hi,
I'm working on the forecast library and am using the auto.arima function
with some dummy data

Fit<-auto.arima(data_ts, trace=TRUE)

I understand that trace evaluates alternative models and provides the
corresponding AIC/SIC values with an output that looks like this;

ARIMA(2,1,2) with drift ? ? ? ? : 1278.988
?ARIMA(0,1,0) with drift ? ? ? ? : 1281.161
?ARIMA(1,1,0) with drift ? ? ? ? : 1280.325
?ARIMA(0,1,1) with drift ? ? ? ? : 1278.2
?ARIMA(1,1,1) with drift ? ? ? ? : 1280.229
?ARIMA(0,1,2) with drift ? ? ? ? : 1278.824
?ARIMA(1,1,2) with drift ? ? ? ? : 1281.911
?ARIMA(0,1,1) ? ? ? ? ? ? ? ? ? ?: 1277.73
?ARIMA(1,1,1) ? ? ? ? ? ? ? ? ? ?: 1279.804
?ARIMA(0,1,0) ? ? ? ? ? ? ? ? ? ?: 1281.265
?ARIMA(0,1,2) ? ? ? ? ? ? ? ? ? ?: 1278.626
?ARIMA(1,1,2) ? ? ? ? ? ? ? ? ? ?: 1281.64

?Best model: ARIMA(0,1,1)

Unfortunately, Trace is not an object and i'm unable to read it in to a
data
frame to do further diagnosis.

Can someone please help me with how i can get this output into a data
frame.
I'm unable to do data.frame since i can't coerce an ARIMA object into
this
class.

Sorry if this has been addressed before. I just think there may be some
data
series where running a closely competing alternative model may be useful

Thanks,
AK

-----Original Message-----
From: r-sig-finance-bounces at r-project.org
[mailto:r-sig-finance-bounces at r-project.org] On Behalf Of Daniel Cegielka
Sent: 26 September 2011 17:20
To: chrisbird
Cc: r-sig-finance at r-project.org
Subject: Re: [R-SIG-Finance] Filtering dates/times from zoo/xts series

2011/9/26 chrisbird <chris at chrisbird.com>

Thanks Brian,

I did try using the ['T09:00/T21:00'] method for extraction but it did
not
return anything - I will reinvestigate this and see if I can get it
working.

It works... you are sure that the data was ok? Very strange that you have
received nothing...

d<-xts(1:25, Sys.time() + 1:25)> d ? ? ? ? ? ? ? ? ? ?[,1]
2011-09-26 13:26:55 ? ?1
2011-09-26 13:26:56 ? ?2
2011-09-26 13:26:57 ? ?3
2011-09-26 13:26:58 ? ?4
2011-09-26 13:26:59 ? ?5
2011-09-26 13:27:00 ? ?6
2011-09-26 13:27:01 ? ?7
2011-09-26 13:27:02 ? ?8
2011-09-26 13:27:03 ? ?9
2011-09-26 13:27:04 ? 10
2011-09-26 13:27:05 ? 11
2011-09-26 13:27:06 ? 12
2011-09-26 13:27:07 ? 13
2011-09-26 13:27:08 ? 14
2011-09-26 13:27:09 ? 15
2011-09-26 13:27:10 ? 16
2011-09-26 13:27:11 ? 17
2011-09-26 13:27:12 ? 18
2011-09-26 13:27:13 ? 19
2011-09-26 13:27:14 ? 20
2011-09-26 13:27:15 ? 21
2011-09-26 13:27:16 ? 22
2011-09-26 13:27:17 ? 23
2011-09-26 13:27:18 ? 24
2011-09-26 13:27:19 ? 25> d["2011-09-26 13:27:00/2011-09-26 13:27:10"]
? ? ? ? ? ? ? ? ? [,1]
2011-09-26 13:27:00 ? ?6
2011-09-26 13:27:01 ? ?7
2011-09-26 13:27:02 ? ?8
2011-09-26 13:27:03 ? ?9
2011-09-26 13:27:04 ? 10
2011-09-26 13:27:05 ? 11
2011-09-26 13:27:06 ? 12
2011-09-26 13:27:07 ? 13
2011-09-26 13:27:08 ? 14
2011-09-26 13:27:09 ? 15
2011-09-26 13:27:10 ? 16> d["T13:27:00/T13:27:10"] ? ? ? ? ? ? ? ? ?
?[,1]
2011-09-26 13:27:00 ? ?6
2011-09-26 13:27:01 ? ?7
2011-09-26 13:27:02 ? ?8
2011-09-26 13:27:03 ? ?9
2011-09-26 13:27:04 ? 10
2011-09-26 13:27:05 ? 11
2011-09-26 13:27:06 ? 12
2011-09-26 13:27:07 ? 13
2011-09-26 13:27:08 ? 14
2011-09-26 13:27:09 ? 15
2011-09-26 13:27:10 ? 16

The processing is not to remove non-trading days/holidays - I do that
elsewhere. I'm doing processing on some complex strategies which use
some
instruments which trade a lot, but not everyday. I only wish to process
the
data from liquid days and strip out the less liquid data.

It's quite a sophisticated approach to data. Probably when you filter
using
the time you would have to count the number of observations and does not
bind data below a certain level. I have no idea how to do it in an
elegant
way.

best regards,
daniel

I will certainly investigate quantstrat.

Thanks,

Chris.

--
View this message in context:

http://r.789695.n4.nabble.com/Filtering-dates-times-from-zoo-xts-series-tp38
42937p3843408.html
Sent from the Rmetrics mailing list archive at Nabble.com.

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions
should go.

? ? ? ?[[alternative HTML version deleted]]

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions
should go.

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions
should go.

--
Jeffrey Ryan
jeffrey.ryan at lemnica.com

www.lemnica.com
www.esotericR.com

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions
should go.